Boffins boost battery life with underclocked Wi-Fi

Scientists from the University of Michigan have devised a power management system that can greatly improve the lifespan of radio devices such as smartphones and laptops. Engineering professor Kang Shin and doctoral student Xinyu Zhang have developed a technique for saving the battery life of Wi-Fi devices by clocking down the …

[Arctic fox]

Potentially good news for those who have their wi-fi on most or...........

..............all of the time on a regular basis. I only wonder how common that usage model is? Or have I misunderstood something here? (Which is of course entirely possible)

[John Smith 19]

Which begs the question

Are these tags *already* in the WiFi standard (IE possible for years but no mfg *bothered* to use them) or new ones that need a base station software update as well.

The other question is how many WiFi devices *have* the hardware support necessary to power down further than they do already.

Sound's like a potentially very good thing for WiFi users but it's taken a while, has it no.

[Turtle_Fan]

Cool story bro

Amen to that

[Anonymous Coward]

Hmmm

Not sure this can be achieved with just software upgrades. The largest component of power consumption during "idle listening" is the radio.

How the FW could do something and turn off the radio in time, before the whole frame has been received, without the HW having specific support for indicating this special tag.The bytes are coming in quicker than the speed of the embedded processor core. Perhaps the TIM IE tag is what they are talking about. If so any decent chip already does this, and kills the receive when the TIM does not match.

For the curious, http://linuxwireless.org/en/developers/Documentation/ieee80211/power-savings#The_TIM_information_element

Also there might also be some hidden cost, like excessive lag or dropped packets on the connection.

Will wait to see the paper..

[Peter Gathercole]

I would guess

that most of the time, people use the WiFi on a smartphone in bursts. What I think you could do is to have a delayed power down, so that all the time there is a stream of packets, the radio would remain on, but as soon as there is a gap in the flow of inbound packets destined for that device for more than the delay, it would power down the radio. Generally speaking, people will not be running server-type services on a smartphone (OK, I know that there are exceptions, like uPNP and DLNA), but most things will be initiated from the smartphone.

I suspect that for TCP type services, you could deliberately ignore or even NAK the first packet to force a re-transmit while the radio powers up. Would not work for UDP, icmp or lower level protocols, but UDP services normally have some mechanism for handling lost packets, and I would doubt that many people use layer 1 services on a smartphone.

[Anonymous Coward]

Uapsd

What you describe about bursting data is already part of wifi and is called uapsd. Idle listening as they describe is when the device wakes up for beacons.

Also you describe application layer protocols, radio etc is layer 1+2, and the standard already covers retries.

The guys here seem to specifically talk about under clocking as their contribution. How this would fit in is what the paper should elaborate. I have a sneaky suspicion that they assume the low data rate beacons can be demodulated at a lower clock. Even so, being digital/ dsp, the power saving, relative to the analog radio is insignificant.

[Stephen Moll]

Sounds good but...

I would rather have smarter settings on the phone, for example location based settings. The phone knows where it is all the time. I would like the wifi to activate, when at home, and turn the wifi off when I go out.

[boatman]

location based settings are already available

Have a gander on android market for a app called "llama" (location aware mobile app)

its has quite alot of options you can play with to toggle various phone settings depending upon location and time.

For me it turns on wifi when I get home, turns on my bluetooth for 3 mins when I leave home or work so that if i get in my car it auto connects to my hands free system. turns off bluetooth when car hands free disconnects and no other bluetooth device is connected.

I think the part in this article about this new WiFi method saving 54% of smartphone battery cant be correct, as system monitering of my phone shows that wifi is only responsible for 12% of battery usage. So I assume it will actually save 54% of that 12% used by the wifi thus actually saving only 6% of my total battery usage....

Still, interesting article.

[NogginTheNog]

Wifi Manager...

...on my BlackBerry, a free download widget from the App World, does this. It uses cell site IDs to identify locations, and turns the wifi on the phone on and off appropriately.

I'd be surprised if there wasn't something similar available for Android and/or iPhone?

[Camilla Smythe]

Re: Sounds Rather Good... @ Stephen Moll

"I would rather have smarter settings on the phone, for example location based settings. The phone knows where it is all the time. I would like the wife to activate, when at home, and turn the wife off when I go out."

...

[alan buxey]

..you already can..

well, on Android you can. eg http://www.appbrain.com/app/pl.polidea.y5

[Alan Brown]

or tasker

tasker can do much of the location-aware stuf that llama does plus more. :)

[Tom 7]

2/3 power in idle mode??

Given the max legal transmission power for wifi is 1W this suggests that the receiver is using a couple of watts at least???

What!!!!!

[Mark #255]

It's quite amazing the number of errors you've squashed into that sentence.

Firstly, the power required to receive a transmission is totally uncorrelated with the transmit power.

Secondly, 2/3 of 1W is very definitely not 'a couple of watts'.

Thirdly, the maximum power output is 100 mW. Though I suppose you might be a USAian, where it is 1W.

[rurwin]

WIFE ??

"I would like the wife to activate, when at home, and turn the wife off when I go out."

I'm sure many people would. (Not me though, honest)

[Martin Budden]

V x A = W

"the power consumption is proportional ... to the square of voltage"

Unless my high-school formula is incorrect or I've missed something else (entirely possible), why is he talking about the SQUARE of voltage???